Users Guide

Stacking 263
3 A protocol may simply restart after the failover if neighbors react slowly
enough that they will not normally detect the outage.
The NSF feature enables the management unit in the stack unit to
synchronize the running-config within 60 seconds after a configuration
change has been made. However, if a lot of configuration changes happen
concurrently, NSF uses a back-off mechanism to reduce the load on the
switch. In this case, the management unit in the stack will attempt
resynchronization no more often than once every 120 seconds.
The show nsf command output includes information about when the next
running-config synchronization will occur.
Initiating a Failover
The NSF feature allows the administrator to initiate a failover using the
initiate failover command. This method is preferred over the reload
unit
command as it ensures synchronization of the management unit in the stack
and standby unit.
Initiating a failover reloads the management unit in the stack, triggering the
standby unit to take over. Before the failover, the management unit in the
stack pushes application data and other important information to the standby
unit. Although the handoff is controlled and causes minimal network
disruption, some ephemeral application state is lost, such as pending timers
and other pending internal events. Use the show nsf command to view the
stack checkpoint status prior to reloading a stack member. Do not reload
while a checkpoint operation is in progress.
Always check the stack health before failing over to the standby unit. Use the
show switch stack-ports counters command to verify that the stack ports are
up and no errors are present. Resolve any error conditions prior to failing over
a management unit in the stack. Use the show switch stack-ports stack-path
command to verify the reachability of all stack units. If any units are not
reachable, the stack may split during a failover.
Checkpointing
Switch applications (features) that build up a list of data such as neighbors or
clients can significantly improve their restart behavior by remembering this
data across a warm restart. This data can either be stored persistently, as in